Sophia

Wevolver lists 15 DoF per arm/hand, finger touch sensors, and 600 g hand payload. Wikipedia reports Sophia displayed the ability to create drawings, including portraits, in 2019. No dexterity benchmark or task success rates found.

Wevolver says Sophia mostly operates in 'fully independent AI mode' but mingles with human-written words; Wikipedia says dialogue is generated via a decision tree; Squarespace ethnography describes scripted responses and hybrid human-AI intelligence. No autonomous task execution demos are evidenced.

Wevolver lists 83 total DoF including mobile base and three rolling base options. Wikipedia says Sophia was upgraded with functional legs and ability to walk around January 2018. No speed, stability, or fall data found.

Sensors include RealSense, eye cameras, face detection, body tracking, audio localization and IMU. Wikipedia says computer vision processes input from eye cameras to follow faces, sustain eye contact, and recognize individuals. A non-authoritative Scribd presentation claims 93% emotion recognition accuracy in one study.

Sophia is designed for social interaction with life-like facial expressions, over 60 facial expressions per CNBC, speech synthesis, and many public interviews and events. She has received Saudi citizenship and UNDP Innovation Champion title. However, dialogue is largely scripted via decision tree.

No runtime, MTBF, or reliability metrics found. An ethnography by an insider (Squarespace) states Sophia is far from stable, consistent, and operationally reliable, fragile and apt to malfunction.

Human-sized (167 cm, 20 kg) with wheeled base and intended for nursing homes, events, retail, customer service, therapy and education. Frubber skin and facial expressions support human-robot interaction. No safety certification or formal human-environment fit testing found.